You are on page 1of 7

Renewable Energy 90 (2016) 513e519

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Comparison of Photovoltaic plant power production prediction


methods using a large measured dataset
G. Graditi*, S. Ferlito, G. Adinolfi
ENEA Portici Research Center, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Nowadays the estimation of power production yield by stand-alone and grid-connected Photovoltaic
Received 24 August 2015 (PV) plants is crucial for technical and economic feasibility design analyses. The main goal is to overcome
Received in revised form renewables unpredictability by properly estimating the power production and by suitably balancing
14 November 2015
generation and consumption. In this context, many methods can be applied to forecast renewables
Accepted 5 January 2016
energy production. The scope of this paper is a comparative analysis of three different methods to es-
Available online 18 January 2016
timate the power production of a preexisting PV plant. It is installed at ENEA Research Centre located in
Portici (South Italy) and it is integrated in a Micro Grid (MG) configuration. In detail a phenomenological
Keywords:
Artificial neural network
model proposed by Sandia National Laboratories and two statistical learning models, a Multi-Layer
Estimation Perceptron (MLP) Neural Network and a Regression approach, are compared. These models are deeply
Genetic algorithms different also in terms of required input data and parameters. In detail, phenomenological model
MLP application requires the availability of design parameters and technical devices specifications. Statistical
Photovoltaic power production machine learning models need, however, input variable previously acquired datasets. The a-Si/mc-Si PV
Regression analysis plant, installed at Portici, represents an adequate case study for the three models comparison, as both
design and acquired data are available. In fact, the plant was designed at the ENEA Research Centre so
this makes possible the knowledge of the design parameters and, being a part of the MG, its data are
continuously acquired and transmitted to other network devices. Obtained results demonstrate more
accurate power predictions can be reached by statistical machine learning approaches. The main novelty
of the paper consists in the optimization of the considered models by the appropriate identification of
the minimum and more representative training dataset. Authors underline the unnecessary use of
thousands samples by suitably selecting the dataset size and samples by means of a Genetic Algorithm.
The optimization strategy effectiveness is verified comparing the prediction performances obtained
employing the optimal dataset with those obtained with a randomly chosen dataset. In this scenario,
Genetic Algorithm strategy represents a successful approach to the suitable identification of statistical
models datasets.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction high levels performances in terms of both efficiency and reliability,


so representing promising solutions to the Zero Energy Balance
1.1. Motivation and approach application and realization. The price reduction and the provided
user-friendly cable solutions have contributed to these limitless
In the last years, global demand for electric energy is consider- sources diffusion in civil and industrial contexts [2]. These systems
ably increased, so requiring the identification of new energetic are constituted by large amount of devices: PV generators [3],
solutions. In this scenario, integration of renewable energies, in Distributed Maximum Power Point Tracking (DMPPT) converters
preexisting energy systems represents an important task to be [4,5], inverters, storage systems, grid interface devices, etc. Proper
considered [1]. At present, PV and wind technologies have reached and efficient operating modes are assured only by conveniently
sized and matched components. To achieve this aim, a previous
accurate model development is an essential task.
An aspect influencing renewable energy sources widespread is
* Corresponding author.
their intrinsic unpredictability. In fact, their power production
E-mail address: giorgio.graditi@enea.it (G. Graditi).

http://dx.doi.org/10.1016/j.renene.2016.01.027
0960-1481/© 2016 Elsevier Ltd. All rights reserved.
514 G. Graditi et al. / Renewable Energy 90 (2016) 513e519

size a stand alone PV plant, using a minimum input dataset. An


Nomenclature ANFIS was applied in Ref. [11] to model the different devices
constituting a PV power system and its output signals. In Refs.
ANFIS Adaptive Neuro-Fuzzy Inference System [12,13], a linear regression model and an ANN were applied to es-
ANN Artificial neural network timate daily global solar radiation. A hybrid model can represent a
DAS Data Acquisition System further approach [14]. It combines different models to overcome
DMPPT Distributed Maximum Power Point Tracking limitations characterizing one single technique. In addition,
GAs Genetic Algorithms “ensemble” methods [15] build predictive models by integrating
GIpoa [W/m2] Global Irradiation on plane of array multiple strategies in order to improve the overall prediction per-
GIcsk [W/m2] Global Clear Sky Irradiation formance. In literature, terms “forecast” and “estimate” are often
LS Least Square interchangeably used. More correctly, the term “forecast” should be
MG Micro Grid adopted to provide an expectation of future values of considered
MLP Multi-Layer Perceptron variables based on previously known/acquired data. On the other
MLR Multiple Linear Regression hand, the term “estimate” means to represent the relation between
PV Photovoltaic output and inputs, at the same timestamp. In this paper, the aim is
RES Renewable Energy Sources to provide an “estimate” for an output variable. The models provide
SAM System Advisor Model a single output value without a prediction horizon knowing inputs
SVM Support Vector Machine at the same timestamp.
Tamb [ C] Ambient temperature
Tm [ C] PV module temperature 1.3. Innovative contribution
Wspeed [m/s] Wind speed
In this paper, the PV power production of a 1 kWp experimental
micro-morph silicon PV plant is estimated. It is an experimental
facility installed at the ENEA - Italian National agency for new
depends on continuously changing and not predictable variables, technologies, Energy and sustainable economic development -
such as irradiance, temperature, wind speed and shadows. As a Research Center of Portici located in the South Italy. Authors main
consequence, the introduction of not programmable energy sour- attention is focused on the comparative analysis between a
ces in energy supply systems presume the availability of suitable phenomenological and two statistical learning techniques taking
techniques and tools to accurately predict renewables productions advantage of meteorological, thermal and electrical in situ acquired
in order to correctly schedule them with other energy sources for data. The innovative aspect consists in the optimization of ANN and
generation-load balancer requirements. In this scenario, the iden- Regression models training datasets. These models need to analyze
tification of methods for a meticulous estimation of alternatives previously acquired data to “learn” the input/output relationship.
productions under realistic operation conditions [6,7] constitutes a Such approach is advantageous in many cases, especially when data
valid solution to increase these sources integration, thereby are incomplete or unknown or, as in MG, PV plants are continuously
allowing an accurate economic and technical assessment. monitored. In this paper, the attention is also focused on statistical
learning models optimization. The optimization target not only
consists in the training dataset minimization, but also in the
1.2. Literature review identification of the most representative and meaningful set of data
to train the ANN and the Regression model.
In literature, many studies on forecasting and estimation
methods are available. In detail, the specific literature on PV plant 1.4. Paper organization
power production estimation presents three different types of
models: phenomenological, stochastic/statistical learning and The paper is organized as follows. In section II, the experimental
hybrid ones. PV plant is described. In section III, the different adopted ap-
Deterministic approaches based on physical phenomena try to proaches to estimate PV power production are presented. In section
predict PV plant output by considering the electrical model of the IV, metrics to evaluate models performances are illustrated.
PV devices constituting the plant. Such kind of models are usually Simulation results are reported and discussed in section V. Finally,
implemented in commercial software such as PVSyst, System conclusions and proposal for future works are reported.
Advisor Model (SAM) and so on. For example in Refs. [8], a deter-
ministic approach was used to model electrical, thermal, and op- 2. Experimental PV plant description
tical characteristics of PV modules. By using hourly solar resource
and meteorological data, the model has been validated for different In this paper, a preexisting PV plant installed at the ENEA
modules types. Other approaches are the statistical and machine Research Center is considered. It consists in 28 amorphous and m-
learning ones, such as: ANN, Support Vector Machine (SVM), crystalline silicon (a-Si/mc-Si) modules [16,17], divided in 14 strings
Multiple Linear Regression (MLR), Adaptive Neuro-Fuzzy Inference of two series connected generators. They are 37.5 Wp micromorph
System (ANFIS). These methods operate without any a priori silicon commercial PV generators connected to a commercial 1 kWp
knowledge of the system under consideration. They try to “un- output nominal power inverter. The PV modules are South oriented
derstand” the relation between inputs and outputs by adequately and installed with a 20 tilt angle on a roof structure. PV plant and
analyzing a dataset containing acquired input and output variables DAS details are reported in Fig. 1. This PV plant represents a suitable
collection. Statistical learning algorithms have many advantages. In case study for the models comparative analysis since both its
fact, starting from examples, they are able to learn from them, and design and acquired data are available. In detail, the carried out
they can work also in presence of incomplete data. Once trained, analysis uses a dataset consisting of meteorological, thermal and
they are able to generalize and to provide predictions. Their fea- electrical data acquired from 2006 to 2012 and sampled every ten
tures make them suitable to be used in different contexts. For minutes. This dataset is constituted by the global radiation on plane
example, in Refs. [9,10], an adaptive ANN was used to model and of array GIpoa, the ambient temperature Tamb, the wind speed
G. Graditi et al. / Renewable Energy 90 (2016) 513e519 515

Fig. 1. PV plant and DAS at ENEA Portici Research Center.

Wspeed, and the PV module temperature Tm. In addition, produced plant losses by means of a “derating” factor calculated with Loss
DC and AC power are taken into account. Calculator software provided by NREL [18] and included in
PVWatts® calculator. This model can predict PV plant power pro-
3. Methodology duction even before the PV system is physically realized, consid-
ering only systems information and configuration.
3.1. Input variables selection The Schokley-Sandia model is available as a set of Matlab
functions, collectively labeled as PV_LIB Toolbox. A schematic data
The correct application of the above mentioned models needs a flow diagram of the proposed model is depicted in Fig. 2. In this
preliminary task represented by input variables selection. The scheme we have adopted the same nomenclature of Matlab
identification of variables which strongly affect the system power
production can be carried out by a correlation analysis. The corre-
lation between the PV plant power production, Pac, and a set of
available measured variables is performed. The theoretic clear sky
radiation GIcsk is also taken into account. Pearson correlation values
(R), obtained considering the year 2007, are summarized in Table 1.
The correlation analysis shows that GIpoa, the module temperature
Tm and GIcsk are highly correlated variables with Pac.
Similar results are obtained repeating correlation analysis for
other years, but years 2006 and 2007 show a slightly higher cor-
relation mainly thanks to a preliminary calibration procedure of
devices and solar sensors used for the long-period monitoring of
the PV plant under test. Since only few months data were acquired
during 2006, the year 2007 is selected as training year for ANN and
Regression models.

3.2. Schokley-Sandia Model

The Schokley-Sandia Model, proposed by Sandia National Lab-


oratories [18], estimates PV plants performance by means of
weather data, devices specifications and parameters using a
defined set of equations. This modeling approach requires a great
amount of details: PV generator thermal and electrical parameters,
array configuration, etc. Moreover, this model takes into account all

Table 1
Correlation analysis results for year 2007.

Variable (V) R-Pearson correlation (V vs Pac)

GIpoa 0.99
GIcsk 0.78
Wspeed 0.19
Tamb 0.45
Tm 0.97
Fig. 2. Schematic data flow diagram of the model developed by Sandia.
516 G. Graditi et al. / Renewable Energy 90 (2016) 513e519

functions utilized in PV_LIB Toolbox. In detail, GIpoa, Wspeed and shown. In this study, the robust fitting version of the method is
Tamb represent the input variables for the considered model. implemented with the aim to mitigate the high outliers sensitivity
characterizing LS estimation. In robust regression, weights are
assigned to each data point; this is done automatically and itera-
3.3. Regression model tively by means of a method consisting of iteratively reweighting
LS. At the first iteration, equal weights are assigned to the model
Another proposed model is a multiple linear regression. In a coefficients calculated by the traditional LS. In subsequent itera-
regression model, an equation describing the statistical relation- tions, weights are recalculated in such a way that lower weights are
ship between one or more predictors and the response variable is assigned to points farther from model predictions. This procedure
derived with the aim to obtain new observations. A multiple linear goes on until coefficients values converge within a specified
regression model can be expressed as follows tolerance.
All common tests have been performed (normal probability plot
yi ¼ b0 þ b1 Xi1 þ b2 Xi2 þ …/ þ bp Xip þ εi
(1) of residuals, variance inflation factors, etc.) to check if LS regression
i ¼ 1; …:n; j ¼ 1; …; p
produces unbiased coefficient estimates with the minimum
variance.
where yi is the i-th response, bk is the k-th coefficient, b0 is the
constant term. Xij is the i-th observation on the j-th predictor var-
3.4. Artificial neural network model
iable, εi is the i-th random error term. In general, a linear regression
model can be expressed as in (2):
ANNs were originally proposed as simplified models of biolog-
XK   ical neural networks. As their biological counterpart, ANN can learn
yi ¼ b0 þ b f Xi1 ; Xi2 ; Xi3 ;……::Xip þ εi
k¼1 k k (2) from examples. For this reason, they are becoming useful as alter-
i ¼ 1;…:n native to conventional techniques to solve a wide variety of com-
plex, non-linear, and even non-stationary problems. ANNs are
where f(-) is a scalar-valued function of the independent variables
widely used in the field of the renewable energies to forecast or
Xij. It is worth noting that in “linear” regression the response vari-
estimate energy production or other variables related to renew-
able, y, is a linear function of the coefficients, bk. In our case, the aim
ables (e.g. solar radiation, wind power, energy demand, etc.)
is to obtain a relation between Pac and the predictors variables
[19,20].
represented by GIpoa, GIcsk and Tm. The form of the adopted
In this paper, a Multi-Layer-Perceptron (MLP) ANN is selected in
regression equation is reported in (3):
order to estimate the PV power production Pac considering GIpoa,
pac ¼ b0 þ b1 *GIpoa þ b2 *GIcsk þ b3 *Tm (3) GIcsk. and Tm as input variables. The MLP is a particular architecture
of ANN, where base units are arranged in layers with only forward
where b3 takes into account of the influence of Tm on the Pac. connections to units in subsequent layers [7]. Such type of ANNs
Regression coefficients are estimated to minimize the mean can properly approximate any non-linear functions and it is usually
squared difference between the prediction vector and the true employed in regression (estimate an output variable using pre-
(measured) response vector. This method is known as the Least dictors variables) problems. In Fig. 4 a diagram of the employed
Squares (LS) method. In Fig. 3, the Regression model flowchart is MLP is shown.

3.5. ANN and regression models dataset optimization

ANN and Regression model used to estimate the power pro-


duction from renewable generators can find a real world significant
applicability in Micro Grids (MG) and Smart Grids (SG) scenarios.
The installed capacity of PV systems is continuously increased at a
much faster rate than the development of electric network, in
terms of grid codes, advanced devices and smart interfaces, etc,
needed to effectively and efficiently manage high penetration of PV
within the distribution system and to allow an active participation
of PV generation in grid control. In fact, in SG contexts, character-
ized by distributed generation from RES and electric storage sys-
tems, usually PV plants are continuously monitored to input/output
data transmission to other network systems to allow the activation
of “smart” management and control strategies. The aim is to
guarantee a suitable reliability and efficiency of the electric grid
operation. Moreover in SG, this data exchange is useful for suitable
energy scheduling among generation/consumption systems and for
safety and diagnostic reasons during the grid operation [21]. So
large datasets are available and they can be used to estimate PV
plants power production without any other information about the
PV plant technical details. In case of dataset constituted by thou-
sands samples, it could be useful to suitably extract an ‘optimal’
representative subset characterized by good generalization capa-
bilities and avoiding over fitting. In this paper a GA [22] is used to
identify minimum and meaningful datasets.
GAs are global minimum search techniques particularly useful
Fig. 3. Regression model flowchart. in optimization process where the objective function is
G. Graditi et al. / Renewable Energy 90 (2016) 513e519 517

Fig. 4. MLP schematic block.

discontinuous, non-differentiable or highly nonlinear. The first step penalty value reaches its minimum after few generations (4)
in the optimization process is to create an initial population by a remaining almost the same for the subsequent 50 generations. This
random selection of values. The fitness function for the GA has been process goes on until a stopping condition is reached: Stall Time (T),
chosen as the normalized PV plant production Root Mean Square Stall Generation (G), Time and Generation. In case of the ANN, GA
Error (nRMSE). Individuals gaining higher nRMSE are chosen as stops being Stall Generation criterion reached, as reported in Fig.6b.
elite, thereafter applying crossover and mutations operators sub- A number of 54 generations are used to identify the minimum and
sequent populations are created. Authors decided to apply elitism, meaningful dataset characterizing the developed ANN. In a similar
two-point crossover and mutation. In detail, chosen crossover and manner, in case of Regression model, the applied GA reaches the
mutation rates are 90% and 10% respectively. A flowchart of the Stall Generation stopping criterion with a number of 57
applied GA is showed in Fig. 5. GA performance evolves as number generations.
of generation increases, as depicted in Fig. 6a, where the best

4. Metrics to evaluate model prediction accuracy

The following metrics (4)e(6) are adopted to evaluate and


compare model predictions [23]:

 normalized Root Mean Square Percentage Error (nRMSE%)


vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u n
1u 1X
nRMSE% ¼ 100* t ðei Þ2 (4)
y n
i¼1

Fig. 6. (a) ANN optimization: GA performance vs. generation number graph, (b) GA
Fig. 5. GA flowchart. stopping criteria graph.
518 G. Graditi et al. / Renewable Energy 90 (2016) 513e519

Table 2 models is fulfilled. The Schokley-Sandia model is our reference to


Prediction accuracy results. perform a comparative analysis of the proposed methods. In more
Model details, Robust Regression LS fit procedure, calculated using data
Year Sandia Regression ANN
concerning year 2007, produces the regression equation reported in
(7):
2006 nRMSE ¼ 14.12% nRMSE ¼ 7.96% nRMSE ¼ 7.51%
RMSE ¼ 55.18 [W] RMSE ¼ 32.13 [W] RMSE ¼ 30.43 [W]
MAPE ¼ 11.83% MAPE ¼ 7.28% MAPE ¼ 6.31%
R2 ¼ 0.9821 R2 ¼ 0.9777 R2 ¼ 0.9800 Pac ¼ 32:142 þ 0:80438*GIpoa þ 0:07646*GHIcsk
2007 nRMSE ¼ 9.15% nRMSE ¼ 5.03% nRMSE ¼ 4.46%
þ 0:30053*Tm (7)
RMSE ¼ 34.78 [W] RMSE ¼ 19.71 [W] RMSE ¼ 17.56 [W]
MAPE ¼ 9.13% MAPE ¼ 5.17% MAPE ¼ 4.40%
Concerning the designed MLP, its main parameters are:
R2 ¼ 0.9858 R2 ¼ 0.9906 R2 ¼ 0.9926
2008 nRMSE ¼ 7.79% nRMSE ¼ 5.75% nRMSE ¼ 5.33%
RMSE ¼ 28.11 [W] RMSE ¼ 21.28 [W] RMSE ¼ 19.78 [W]  1 hidden layer with 3 neurons;
MAPE ¼ 7.96% MAPE ¼ 5.50% MAPE ¼ 4.85%  LevenbergeMarquardt training algorithm;
R2 ¼ 0.9838 R2 ¼ 0.9877 R2 ¼ 0.9894  80% of whole dataset (year 2007) for training, 10% for validation
2009 nRMSE ¼ 9.04% nRMSE ¼ 5.38% nRMSE ¼ 5.60%
and 10% for test;
RMSE ¼ 35.89 [W] RMSE ¼ 22.12 [W] RMSE ¼ 23.28 [W]
MAPE ¼ 9.14% MAPE ¼ 5.06% MAPE ¼ 5.70%  Log-sigmoid transfer function for the hidden layer;
R2 ¼ 0.9845 R2 ¼ 0.9884 R2 ¼ 0.9871  Linear transfer function for the output layer.
2010 nRMSE ¼ 12.24% nRMSE ¼ 8.37% nRMSE ¼ 8.46%
RMSE ¼ 44.26 [W] RMSE ¼ 31.16 [W] RMSE ¼ 31.71 [W]
Preliminary results, shown in Table 2, underline the different
MAPE ¼ 9.68% MAPE ¼ 7.11% MAPE ¼ 7.63%
R2 ¼ 0.9650 R2 ¼ 0.9758 R2 ¼ 0.9749
performances achievable with the three considered models. Both
2011 nRMSE ¼ 9.99% nRMSE ¼ 7.78% nRMSE ¼ 7.19% in terms of R2 and nRMSE%, the ANN model represents the one
RMSE ¼ 37.41 [W] RMSE ¼ 29.91 [W] RMSE ¼ 27.72 [W] that best fits the experimental data. ANN reaches the best results,
MAPE ¼ 11.86% MAPE ¼ 7.98% MAPE ¼ 7.38% especially for the year 2007, with a nRMSE% value of 4.46%.
R2 ¼ 0.9713 R2 ¼ 0.9757 R2 ¼ 0.9791
Considering the same year, nRMSE% values of 5.03% and 9.15% are
2012 nRMSE ¼ 8.36% nRMSE ¼ 8.85% nRMSE ¼ 8.10%
RMSE ¼ 30.09 [W] RMSE ¼ 32.84 [W] RMSE ¼ 29.99 [W] obtained for Regression and Schokley-Sandia models, respec-
MAPE ¼ 0.25% MAPE ¼ 8.99% MAPE ¼ 8.13% tively. Worst results in terms of prediction accuracy regards the
R2 ¼ 0.9780 R2 ¼ 0.9664 R2 ¼ 0.9720 year 2010: ANN is characterized by a nRMSE% of 8.46%, Schokley-
nRMSE%_AVG 10.09 7.01 6.66
Sandia model reaches a value of 12.24% and Regression a value of
R2_AVG 0.978 0.980 0.982
8.37%. It is due to a usual and long period of variable and overcast
days. In addition, the case study continues comparing the opti-
 Root Mean Square Error (RMSE) mized Regression and ANN methods with the not optimized
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ones.
u n
u1 X The optimization procedure underlines the unnecessary use of
RMSE ¼ t ðei Þ2 (5) thousands samples or different months samples to identify a
n
i¼1
minimum and meaningful dataset for ANN and Regression ap-
proaches. Both in case of ANN and Regression, the adopted GA se-
lects only two days samples. In details, the 10th and 23rd of April
 Mean Absolute Percent Error (MAPE)
are chosen for ANN, while the 17th and 23rd of April are considered
for the Regression model.
1Xn
jei j
MAPE ¼ 100* (6) Comparative analysis results are shown in Table 3. For reason of
n yi
i¼1 clarity, only nRMSE values are reported. It is worth noting that
where:
nRMSE values obtained for the optimized models are comparable
with those calculated acquiring information for a whole year. The
ei ¼ (Pac  Pac )
measured,i predicted,i suitable identification of optima days represents a crucial task. In
n ¼ number of samples the last column of Table 3, samples referring to two randomly
yi ¼ Pac chosen days are considered and provided to ANN model as training
measured,i
set. Obtained results point out the random choice of a minimum
dataset does not represent a successful strategy. In fact, samples
5. Results evaluation and discussion must be chosen in order to obtain not only minimum, but also a
representative dataset able to train the ANN. In this scenario, GAs
In the next, the comparative analysis among the above reported can provide advantageous solutions.

Table 3
Comparative analysis nRMSE results.

Model

Year Sandia Regression Optimized regression ANN Optimized ANN Random ANN

2006 14.12% 7.96% 7.72% 7.51% 7.89% 7.91%


2007 9.15% 5.03% 5.00% 4.46% 4.62% 4.84%
2008 7.79% 5.75% 5.77% 5.33% 5.31% 5.62%
2009 9.04% 5.38% 5.38% 5.60% 5.64% 5.73%
2010 12.24% 8.37% 8.54% 8.46% 8.15% 9.27%
2011 9.99% 7.78% 7.94% 7.19% 7.15% 7.28%
2012 8.36% 8.85% 8.79% 8.10% 7.81% 8.40%
G. Graditi et al. / Renewable Energy 90 (2016) 513e519 519

6. Conclusion forecasting tool on short- term horizon, in: 2011 19th Mediterranean Con-
ference on Control & Automation (MED), 2011, pp. 1265e1270.
[8] D.L. King, W.E. Boyson, J. a Kratochvill, Photovoltaic Array Performance Model,
The accurate estimation of the power production is very December, 2004, pp. 1e43.
important for the operations planning of electric power systems. In [9] T. Hiyama, E. Karatepe, Investigation of ANN performance for tracking the
this paper, two statistical learning models and a physical/ optimum points of PV module under partially shaded conditions, in: 2010
Conference Proceedings IPEC, 2010, pp. 1186e1191.
phenomenological one are compared. Performance results show [10] D.L. King, W.E. Boyson, J. a Kratochvill, Photovoltaic Array Performance Model,
that more accurate predictions can be obtained by Regression and December, 2004, pp. 1e43.
MLP models as compared to the classical reference method. Further [11] A. Mellit, Artificial intelligence based-modeling for sizing of a stand-alone
photovoltaic power system: proposition for a new model using neuro-fuzzy
studies identifies the minimum dataset necessary to train ANN and system (ANFIS), in: 2006 3rd International IEEE Conference Intelligent Sys-
regression models by means of a GA. The minimum and repre- tems, 2006, pp. 606e611.
sentative dataset is constituted by only two days samples both for [12] M. Bocco, E. Willington, M. Arias, Comparison of regression and neural net-
works models to estimate solar radiation, Chil. J. Agric. Res. 70 (3) (2010)
ANN and Regression analysis. The GA solutions effectiveness is 428e435.
verified comparing ANN performances in case of optimized and [13] P.G. Nikhil, D. Subhakar, Approaches for developing a regression model for
random chosen training datasets. sizing a stand-alone photovoltaic system, IEEE J. Photovolt. 5 (1) (Jan. 2015)
250e257.
Future studies are going to analyze optima datasets to identify [14] Y.K. Wu, C.R. Chen, H.A. Rahman, A novel hybrid model for short-term fore-
suitable merit factors. The idea is to classify days respect to the casting in PV power generation, Int. J. Photoenergy 2014 (2014).
optima selected ones on the base of specific indices. In this way, [15] J. Yokoyama, Short term load forecasting improved by ensemble and its
variations, in: 2012 IEEE Power and Energy Society General Meeting, 2012, pp.
acquiring two days data in every season or month of the year, it is
1e6.
possible to correlate these samples with the reference ones. This [16] F. Meillaud, A. Billet, C. Battaglia, et al., Latest developments of high-efficiency
analysis could permit a reliable PV power prediction only acquiring micromorph tandem silicon solar cells implementing innovative substrate
small samples. materials and improved cell design, IEEE J. Photovolt. 2 (3) (Jul. 2012)
236e240.
[17] M. Boccard, P. Cuony, C. Battaglia, et al., Nanometer- and micrometer-scale
References texturing for high-efficiency micromorph thin-film silicon solar cells, IEEE J.
Photovolt. 2 (2) (Apr. 2012) 83e87.
[1] C. Paravalos, E. Koutroulis, V. Samoladas, T. Kerekes, D. Sera, R. Teodorescu, [18] W. Andrews, J. S. Stein, C. Hansen, D. Riley, C. Consulting, and S. N. Labora-
Optimal design of photovoltaic systems using high time-resolution meteo- tories, Introduction to the Open Source PV LIB for Python Photovoltaic System
rological data, IEEE Trans. Ind. Inf. 10 (4) (Nov. 2014) 2270e2279. Modelling Package, pp.1e5.
[2] G. Graditi, M.L. Di Silvestre, R. Gallea, E. Riva Sanseverino, Heuristic-based [19] D.M. Riley, G.K. Venayagamoorthy, Comparison of a recurrent neural network
shiftable loads optimal management in smart micro-grids, IEEE Trans. Ind. Inf. PV system model with a traditional component-based PV system model, in:
11 (1) (Feb. 2015) 271e280. Conf. Rec. IEEE Photovolt. Spec. Conf., No. June, 2011, pp. 002426e002431.
[3] D.L. King, W.E. Boyson, J.A. Kratochvill, Photovoltaic Array Performance Model, [20] N. Kassim, S.I. Sulaiman, Z. Othman, I. Musirin, Harmony search-based opti-
Snadia Report, 2004-3535, pp. 1e43. mization of artificial neural network for predicting AC power from a photo-
[4] G. Adinolfi, N. Femia, G. Petrone, G. Spagnuolo, M. Vitelli, Design of dc/dc voltaic system, in: 2014 IEEE 8th International Power Engineering and
converters for DMPPT PV applications based on the concept of energetic ef- Optimization Conference (PEOCO2014), 2014, pp. 504e507.
ficiency, J. Sol. Energy Eng. 132 (2) (May 2010) 021005. [21] M. Braun, et al., Is the distribution grid ready to accept large scale photovoltaic
[5] G. Graditi, G. Adinolfi, Energy performances and reliability evaluation of an deployment? e State of the art, progress and future prospects, Prog. Photo-
optimized DMPPT boost converter, in: 2011 International Conference on Clean volt. Res. Appl. 20 (6) (2011) 681e697.
Electrical Power (ICCEP), 2011, pp. 69e72. [22] R.G. Kamp, H.H.G. Savenije, Optimising training data for ANNs with genetic
[6] G. Graditi, S. Ferlito, G. Adinolfi, G.M. Tina, C. Ventura, Performance estimation algorithms, Hydrol. Earth Syst. Sci. 10 (2006) 603e608.
of a thin-film photovoltaic plant based on an artificial neural network model, [23] M.V. Shcherbakov, A. Brebels, A survey of forecast error measures, World
in: 2014 5th International Renewable Energy Congress (IREC), 2014, pp. 1e6. Appl. Sci. J. Inf. Technol. Mod. Ind. Educ. Soc. 24 (4) (2013) 171e176.
[7] O.E. Dragomir, F. Dragomir, I. Brezeanu, E. Minca, MLP neural network as load

You might also like